Can a robot hold patents? The U.S. thinks not!
As technology becomes more and more innovative, it is becoming easier and easier to imagine that soon artificial intelligence (AI) would be able to make things all by itself. Already, we have seen a rising concern over AI making its own creative works by recognizing patterns in art styles and structures.
So an AI can be an artist, but can it also be an inventor? Let’s find out how this debate came about, how DABUS (i.e. the machine behind this debate) works and what the law has to say about DABUS.
How did the idea of AI inventors gain popularity?
The debate surrounding AI inventors stems from the efforts of law professor, Dr. Ryan Abbott, and the Founder of Imagination Engines Inc, Stephen Thaler. Abbott is responsible for “The Artificial Inventor Project”, which enlisted Thaler to create a machine built simply to invent. This machine is called DABUS, or Device for the Autonomous Bootstrapping of Unified Sentience.
“The Artificial Inventor Project” came into the limelight in 2019 and has since filed patent applications for two of DABUS’s inventions (a food container and a light beacon) in 17 jurisdictions, including the U.S., Europe, the U.K., Australia, Brazil, Saudi Arabia, China and South Africa.
So how does DABUS work?
Imagination Engine’s website explains DABUS as many interconnected neural networks, with each network having interrelated memories of a linguistic, visual or auditory nature. These networks constantly go through cycles of learning and unlearning.
To put it simply, DABUS has been fed general knowledge on a wide variety of topics. It uses this general knowledge to learn and disrupt the links between various topics during its learning process. The disruptions that occur in the learning process are then observed by AI for utility. Ideas with utility are then developed further by going through the same disruption process.
The future of AI inventors in the eyes of the law
The law or the legal system, in general, doesn’t seem to look favorably upon DABUS. Both the U.S. and the U.K. have rejected patent applications filed for DABUS’s inventions because they have not been made by a “person” as the inventor.
In the U.S., the United States Patent and Trademark Office (USPTO) refused to recognize DABUS’s request for a patent for its “failure to identify a natural person as the inventor”. As a result of this, in August 2020, Thaler sued the USPTO in the Eastern District of Virginia, asking the court for a reversal of the USPTO’s decision. However, in September of this year, the court ruled in the USPTO’s favor. The judge presiding over the case told Thaler that his best hope would be a change in U.S. law.
Still, all is not lost for DABUS. One of its inventions has been granted a patent in South Africa. This is because South Africa’s patent process passes every application that fulfills all formalities, regardless of “who” the inventor is. However, the South African patent authority has not stepped in to give a clear statement on where it stands about AI as an inventor.
Similarly, an Australian court has also recognized DABUS’s patent applications, becoming the first country to concede that an inventor need not necessarily be a person. With these two countries offering DABUS a glimmer of hope, we can only wonder whether others will follow their lead in accepting the AI revolution.
Header image courtesy of Freepik